Pre-processsing

Hashtags

DateColumns: + month, year columns

Turn tweets into embedding vectors

Word2Vec model

Clustering

Sentiment analysis of words

Out of 19911 unique words and bigram from the dataset:

11621 (33.61%) are Neutral sentiments 14786 (42.76%) are Positive sentiments 8170 (23.63%) are Negative sentiments

It shows that the Neutral and Positive words have larger domination in the dataset

Custom sentiment analysis of tweets

Out of 77467 tweets from the dataset:

102577(45.6%) are Negative sentiments 102577(9.1%) are Neutral sentiments 20412(45.3%) are Positive sentiments

Data Visualization

Wordcloud

Sentiment Curve for 2022

EDA of results